Identify the consistency of failure mechanisms from accelerated degradation data

Abstract

Accelerated degradation test is an efficient approach to evaluate the reliability of degradation products. An effective accelerated test must guarantee that the failure mechanisms of product under all accelerated stresses are consistent with that under the normal stresses. Otherwise, the reliability of product under the normal stresses cannot be accurately extrapolated from accelerated degradations data. A method of identifying the consistency of failure mechanism is proposed, in which an inverse Gaussian process is used to model accelerated degradation data. First, the acceleration factor constant principle was applied to deduce the changing rules of model parameters with accelerated stresses. Then, the parameter estimates were evaluated from accelerated degradation data by maximum likelihood estimation method. If the parameter estimates obey the changing rules, the failure mechanisms are considered to be consistent, otherwise not. Next, a hypothesis method based on t statistic was developed to verify whether the parameter estimates obey the changing rules or not. Besides, a validation test was designed to assess the performance of the identification method, and a case study was provided to demonstrate the effectiveness and usefulness of the method.